A Game-Changer for Hospital Logistics

Like so many industrial and commercial establishments, hospitals are bemoaning the shortage of skilled workers. The care sector is particularly affected by this. One solution to ease the burden on nursing staff and make the profession more attractive could lie in reconfiguring logistical processes with robots. Autonomous mobile robots (AMRs) have been used to transport high-volume loads such as food, laundry and dumpsters for several years. However, their sphere of activity is limited to areas where there is no foot traffic. But now, with the introduction of the smart transportation robot “RemRob” from Fraunhofer IML, this is set to change. Like R2-D2 from the popular Star Wars saga, the robot is designed to trundle through hospital corridors autonomously and supply the wards with medical materials. 

As part of the research project “5G-Remote Assistance for Robotics” (5G-RemRob), the Fraunhofer Institute for Material Flow and Logistics IML and its industry partners, FACT GmbH, Sick GmbH and the St. Franziskus-Hospital in Münster, are aiming to use artificial intelligence to make existing autonomous transportation robots fit for more widespread use in hospitals, while also ensuring that they are easy to implement. The goal is to gradually enable transportation robots to drive autonomously around different areas of the hospital in order to deliver small amounts of materials, such as defective medical devices, pharmaceuticals or laboratory samples – tasks that are currently carried out by nurses. If robots were to take over these tasks in the future, not only would it help counteract the shortage of nurses, but workflows could also be designed more efficiently and resources could be utilized in a more targeted way, which would ultimately lead to better patient care.

Breathing intelligence into robots

The particular challenge in the 5G-RemRob project was developing a system that could successfully carry out transportation tasks in a chaotic, partially public environment where it would encounter many different groups of untrained people. The technology that the project team used was designed to enable the transportation robot to find independent solutions to specific problems, such as avoiding obstacles or interacting with people. The two-year project, which was funded as part of the 5G.NRW funding competition run by the Ministry of Economic Affairs, Industry, Climate Action and Energy of the State of North Rhine-Westphalia, ended in November 2023 with a pilot phase at the St. Franziskus Hospital in Münster. The research project is based on three developments that each build on each other; however, at the core of them all is the Remote AI Box – the sensor box the robot uses to communicate and transmit and receive data. The AI box combines the sensors and computing hardware the robot needs to navigate through complex environments. 

Hello, operator! Real-time communication between humans and machines

The researchers have developed a remote assistant functionality for cases where the transportation robot cannot independently solve a problem such as an obstacle along its route. This is used to send an error report to a control station overseen by an external operator. The operator can then take control of the robot using XR smart glasses that provide both technical system information and image data from the robot’s surroundings. In addition, the XR technology’s 3D imaging allows the operator to visualize the environment easily, immerse themselves in the robot’s immediate surroundings and help the robot to navigate past the obstacle, so it can then continue with its transportation task autonomously. Fraunhofer IML uses the 5G communication standard to control the robot remotely. This ensures a stable, low-latency wireless network connection and fast, secure, real-time data transfers between different system components. The use of 5G presents another advantage in hospital contexts. If the robot were to use the hospital’s Wi-Fi network, it would be in danger of violating regulations on protecting patient data. 

Continuously learning from experience

The robot is not only connected to a remote user – thanks to neural networks, this person can also train it. The data generated during communication with the operator is processed and stored by an AI algorithm, so the robot learns to deal with different problems over time. In the long term, this will increase the robot’s autonomy during operation, meaning it will require less and less help from its human colleague. The information that individual robots learn can also be transferred to other robots, thanks to the concept of lifelong AI training. With this learning method, the robots can become increasingly autonomous, even in new environments and difficult locations. “The AI Box’s special features include modularity and advanced, integrated AI algorithms for tasks such as image recognition. This means a robot system can not only pinpoint where it is in the environment, but also recognize what is happening there,” says Sebastian Hoose of the Robotics and Cognitive Systems department at Fraunhofer IML. “So the robot doesn’t just see a wheelchair at the hospital exit as an obstacle, but also recognizes the obstacle as a wheelchair and knows that it needs to avoid it.”

A cost-effective, compatible, flexible robot system

With the “RemRob” service robot, hospitals not only receive hardworking helpers for transporting medical materials, but are also given a great degree of flexibility when it comes to choosing the model of their robots. The modular design ensures the Remote AI Box is compatible with robot platforms from different manufacturers. In addition, the number of robots in the fleet can be adapted flexibly, because all the training that a robot has built up at a given point in time can easily be transferred to other robots. Autonomizing logistics transportation processes in hospitals using robots could ease the strain on nursing, logistics and medical technology personnel and allow them to redirect their focus toward their high-value core tasks.

Everyday hospital work: a litmus test for robots

In order to evaluate and demonstrate the actual performance of the 5G-RemRob service robot, Fraunhofer IML and the FACT Group conducted a pilot phase at the St. Franziskus-Hospital in Münster. The aim was to conduct a practical test to demonstrate the technical feasibility of the autonomous transportation solution at the hospital and identify potential challenges. Nursing staff, logistics specialists, medical technicians and building operations managers all brought their expertise to bear in identifying suitable areas of application for the robot. “During the tests, the service robot was given the task of transporting small devices, such as infusion pumps, blood pressure monitors and ECG apparatus, from the wards to the medical technology department for repairs, and vice versa. It also transported boxes of drugs and medications and carried items between wards, such as nightgowns for patients, medications and files, particularly during the night,” says Marcus Hintze, a research scientist in the Health Care Logistics department and head of the 5G-RemRob project. Throughout the tests, the robot transported the goods in locked boxes to prevent theft. 

Challenges in existing hospital infrastructure

During the test phase, the robot undertook its first successful journeys in a hospital ward selected for this purpose. One of the challenges was the existing infrastructure at the St. Franziskus-Hospital. This hospital is a good example of an existing building complex that has been expanded several times over the years. “When using robots in existing buildings, one particular challenge is adapting to the infrastructure,” explains Jan Rasmus, managing director of the FACT Group. “How can the robot handle elevators, roller gates, doors or paved paths? The robot must also be able to solve certain types of problems independently while transporting materials, such as avoiding obstacles or interacting with people. This is important because at some stage, the robots will ultimately need to be integrated into the normal everyday work on the hospital corridors, which often have very heavy foot traffic and may be cluttered with beds or equipment.” The “lifelong training algorithm” is crucial here, because it means that, with the help of AI, the robot can gradually increase its autonomy and capabilities, and so become more adept at navigating within its environment over time. In addition, service robots will need to be able to communicate with motion detectors or automatic doors at the hospital. Smart devices could act as the connecting link here, allowing for wireless communication and interaction. Fire protection requirements constitute a further hurdle. The robot charging station must be located in a separate room with fire alarms, fire-resistant walls and a T30 door.

How economical are the robots?

The “5G-RemRob” research team also investigated the financial viability of using robots in the St. FranziskusHospital. They developed a tool that calculates financial viability once the relevant parameters are entered. This tool will support hospitals in making investment decisions. The calculation compared the manual transportation processes by people with the processes that could potentially be carried out by robots and showed that using robots to transport materials within hospitals can have financial advantages in comparison to transporting the materials manually. Because different hospitals have different building infrastructure, each one has its own very individual requirements for using robots. This means the financial viability must be calculated individually for each hospital

Research on 5G-RemRob continues

After completion of the pilot phase at the St. FranziskusHospital in Münster, discussions with the FACT Group suggested that the capabilities of the sensor box should be expanded in order to make it a more attractive option for other possible areas of application in the service sector beyond medicine. In addition to transportation, the robot could also take on other tasks, e.g., in the area of monitoring and building security, as it passes by the points that need to be checked every day. Among other things, these tasks include fire extinguisher checks (e.g., making sure they are where they should be and have not passed their expiry dates), checking escape routes are clear and checking escape plans, etc. The sensor box does not need any additional hardware components to perform these tasks. Only the software and algorithm would have to be adapted accordingly. The researchers have applied for 250,000 euros of funding from DATIpilot, a German Federal Ministry of Education and Research (BMBF) funding program. The project is due to be launched in September 2024 and will run for a year and a half. “With our support, commercial companies could then take over to develop the concept further and bring it to market maturity and commercialization,” says Marcus Hintze, who is looking forward to discussing this with interested robotics manufacturers. “We believe that hospitals are increasingly going to depend on robots for help. This means robotics will keep on developing to meet these needs and will continue perfecting human-technology interactions in the process,” says Hintze. “As a health care logistics specialist, we believe that care and medical diagnostics should still be a human task, but all non-care services can be automated with robots so that nurses can focus on the patients – a solution that patients are sure to welcome.”

Marcus Hintze

Contact Press / Media

M.Sc. Marcus Hintze

Health Care Logistics

Fraunhofer-Institut für Materialfluss und Logistik IML
Joseph-von-Fraunhofer-Straße 2-4
44227 Dortmund, Deutschland

Phone +49 231/9743-504

Fax +49 231/9743-77-504

Sebastian Hoose, M.Sc.

Contact Press / Media

Sebastian Hoose, M.Sc.

Research fellow - department Robotics and Cognitive Systems

Fraunhofer Institute for Material Flow and Logistics
Joseph-von-Fraunhofer-Straße 2-4
44227 Dortmund

Phone + 49 231 9743-490